diff --git a/src/common/stores/llms/llms.types.ts b/src/common/stores/llms/llms.types.ts index 70f425a8b..07f1a549c 100644 --- a/src/common/stores/llms/llms.types.ts +++ b/src/common/stores/llms/llms.types.ts @@ -31,7 +31,6 @@ export interface DLLM { // hard properties (overwritten on update) contextTokens: DLLMContextTokens; // null: must assume it's unknown maxOutputTokens: DLLMMaxOutputTokens; // null: must assume it's unknown - trainingDataCutoff?: string; // 'Apr 2029' interfaces: DModelInterfaceV1[]; // if set, meaning this is the known and comprehensive set of interfaces benchmark?: { cbaElo?: number, cbaMmlu?: number }; // benchmark values pricing?: DModelPricing; diff --git a/src/modules/llms/llm.client.ts b/src/modules/llms/llm.client.ts index 0198a39b1..b627a80ec 100644 --- a/src/modules/llms/llm.client.ts +++ b/src/modules/llms/llm.client.ts @@ -106,7 +106,6 @@ function _createDLLMFromModelDescription(d: ModelDescriptionSchema, service: DMo // hard properties contextTokens, maxOutputTokens, - trainingDataCutoff: d.trainingDataCutoff, interfaces: d.interfaces?.length ? d.interfaces as DModelInterfaceV1[] : _fallbackInterfaces, benchmark: d.benchmark, // pricing?: ..., // set below, since it needs some adaptation @@ -189,7 +188,6 @@ export function createDLLMUserClone(sourceLlm: DLLM, cloneLabel: string, cloneVa // -- Inherited Hard Properties // contextTokens // maxOutputTokens - // trainingDataCutoff // interfaces // benchmark // pricing diff --git a/src/modules/llms/server/anthropic/anthropic.models.ts b/src/modules/llms/server/anthropic/anthropic.models.ts index b9c50afb5..d99ba83f1 100644 --- a/src/modules/llms/server/anthropic/anthropic.models.ts +++ b/src/modules/llms/server/anthropic/anthropic.models.ts @@ -119,7 +119,6 @@ export const hardcodedAnthropicModels: (ModelDescriptionSchema & { isLegacy?: bo description: 'Most intelligent model with advanced reasoning for complex agentic workflows', contextWindow: 200000, maxCompletionTokens: 64000, - trainingDataCutoff: 'Jan 2025', interfaces: [...IF_4, LLM_IF_ANT_ToolsSearch], parameterSpecs: [...ANT_PAR_WEB, { paramId: 'llmVndAntEffort' }], chatPrice: { input: 5, output: 25, cache: { cType: 'ant-bp', read: 0.50, write: 6.25, duration: 300 } }, @@ -130,7 +129,6 @@ export const hardcodedAnthropicModels: (ModelDescriptionSchema & { isLegacy?: bo description: 'Best model for complex agents and coding, with the highest intelligence across most tasks', contextWindow: 200000, maxCompletionTokens: 64000, - trainingDataCutoff: 'Jan 2025', interfaces: [...IF_4, LLM_IF_ANT_ToolsSearch], parameterSpecs: [...ANT_PAR_WEB, { paramId: 'llmVndAnt1MContext' }, { paramId: 'llmVndAntSkills' }], // Note: Tiered pricing - ≤200K: $3/$15, >200K: $6/$22.50 (with 1M context enabled) @@ -153,7 +151,6 @@ export const hardcodedAnthropicModels: (ModelDescriptionSchema & { isLegacy?: bo description: 'Fastest model with exceptional speed and performance', contextWindow: 200000, maxCompletionTokens: 64000, - trainingDataCutoff: 'Feb 2025', interfaces: IF_4, parameterSpecs: [...ANT_PAR_WEB, { paramId: 'llmVndAntSkills' }], chatPrice: { input: 1, output: 5, cache: { cType: 'ant-bp', read: 0.10, write: 1.25, duration: 300 } }, @@ -166,7 +163,6 @@ export const hardcodedAnthropicModels: (ModelDescriptionSchema & { isLegacy?: bo description: 'Exceptional model for specialized complex tasks requiring advanced reasoning', contextWindow: 200000, maxCompletionTokens: 32000, - trainingDataCutoff: 'Jan 2025', interfaces: IF_4, parameterSpecs: ANT_PAR_WEB, chatPrice: { input: 15, output: 75, cache: { cType: 'ant-bp', read: 1.50, write: 18.75, duration: 300 } }, @@ -181,7 +177,6 @@ export const hardcodedAnthropicModels: (ModelDescriptionSchema & { isLegacy?: bo description: 'Previous flagship model', contextWindow: 200000, maxCompletionTokens: 32000, - trainingDataCutoff: 'Mar 2025', interfaces: IF_4, parameterSpecs: ANT_PAR_WEB, chatPrice: { input: 15, output: 75, cache: { cType: 'ant-bp', read: 1.50, write: 18.75, duration: 300 } }, @@ -193,7 +188,6 @@ export const hardcodedAnthropicModels: (ModelDescriptionSchema & { isLegacy?: bo description: 'High-performance model', contextWindow: 200000, maxCompletionTokens: 64000, - trainingDataCutoff: 'Mar 2025', interfaces: IF_4, parameterSpecs: [...ANT_PAR_WEB, { paramId: 'llmVndAnt1MContext' }], // Note: Tiered pricing - ≤200K: $3/$15, >200K: $6/$22.50 (with 1M context enabled) @@ -218,7 +212,6 @@ export const hardcodedAnthropicModels: (ModelDescriptionSchema & { isLegacy?: bo description: 'High-performance model with early extended thinking. Deprecated October 28, 2025, retiring February 19, 2026.', contextWindow: 200000, maxCompletionTokens: 64000, - trainingDataCutoff: 'Nov 2024', interfaces: IF_4, parameterSpecs: ANT_PAR_WEB, chatPrice: { input: 3, output: 15, cache: { cType: 'ant-bp', read: 0.30, write: 3.75, duration: 300 } }, @@ -236,7 +229,6 @@ export const hardcodedAnthropicModels: (ModelDescriptionSchema & { isLegacy?: bo description: 'Intelligence at blazing speeds. Deprecated December 19, 2025, retiring February 19, 2026.', contextWindow: 200000, maxCompletionTokens: 8192, - trainingDataCutoff: 'Jul 2024', interfaces: IF_4, parameterSpecs: ANT_PAR_WEB, chatPrice: { input: 0.80, output: 4.00, cache: { cType: 'ant-bp', read: 0.08, write: 1.00, duration: 300 } }, @@ -254,7 +246,6 @@ export const hardcodedAnthropicModels: (ModelDescriptionSchema & { isLegacy?: bo description: 'Fast and compact model for near-instant responsiveness', contextWindow: 200000, maxCompletionTokens: 4096, - trainingDataCutoff: 'Aug 2023', interfaces: IF_4, chatPrice: { input: 0.25, output: 1.25, cache: { cType: 'ant-bp', read: 0.03, write: 0.30, duration: 300 } }, benchmark: { cbaElo: 1263, cbaMmlu: 75.1 }, diff --git a/src/modules/llms/server/gemini/gemini.models.ts b/src/modules/llms/server/gemini/gemini.models.ts index 58263bdbe..73b222658 100644 --- a/src/modules/llms/server/gemini/gemini.models.ts +++ b/src/modules/llms/server/gemini/gemini.models.ts @@ -794,7 +794,6 @@ export function geminiModelToModelDescription(geminiModel: GeminiWire_API_Models description: descriptionLong, contextWindow: contextWindow, maxCompletionTokens: outputTokenLimit, - // trainingDataCutoff: knownModel?.trainingDataCutoff, // disabled as we don't get this from Gemini interfaces, parameterSpecs: knownModel?.parameterSpecs, // rateLimits: isGeminiPro ? { reqPerMinute: 60 } : undefined, diff --git a/src/modules/llms/server/llm.server.types.ts b/src/modules/llms/server/llm.server.types.ts index e98f06c50..14bf03cb7 100644 --- a/src/modules/llms/server/llm.server.types.ts +++ b/src/modules/llms/server/llm.server.types.ts @@ -142,7 +142,6 @@ export const ModelDescription_schema = z.object({ parameterSpecs: z.array(ModelParameterSpec_schema).optional(), maxCompletionTokens: z.int().optional(), // initial parameter value for 'llmResponseTokens' // rateLimits: rateLimitsSchema.optional(), - trainingDataCutoff: z.string().optional(), benchmark: BenchmarksScores_schema.optional(), chatPrice: PricingChatGenerate_schema.optional(), hidden: z.boolean().optional(), diff --git a/src/modules/llms/server/models.mappings.ts b/src/modules/llms/server/models.mappings.ts index 6537214b1..b7a2a9c4b 100644 --- a/src/modules/llms/server/models.mappings.ts +++ b/src/modules/llms/server/models.mappings.ts @@ -142,7 +142,6 @@ export function fromManualMapping(mappings: (KnownModel | KnownLink)[], upstream // apply optional fields if (m.parameterSpecs) md.parameterSpecs = m.parameterSpecs; if (m.maxCompletionTokens) md.maxCompletionTokens = m.maxCompletionTokens; - if (m.trainingDataCutoff) md.trainingDataCutoff = m.trainingDataCutoff; if (m.benchmark) md.benchmark = m.benchmark; if (m.chatPrice) md.chatPrice = m.chatPrice; if (m.hidden) md.hidden = true; diff --git a/src/modules/llms/server/openai/models/fastapi.models.ts b/src/modules/llms/server/openai/models/fastapi.models.ts index f0baf94b2..e93c1d4e1 100644 --- a/src/modules/llms/server/openai/models/fastapi.models.ts +++ b/src/modules/llms/server/openai/models/fastapi.models.ts @@ -64,7 +64,6 @@ export function fastAPIModels(models: OpenAIWire_API_Models_List.Model[]): Model interfaces, // parameterSpecs: ... // maxCompletionTokens: ... - // trainingDataCutoff: ... // benchmark: ... // chatPrice, hidden: false, diff --git a/src/modules/llms/server/openai/models/fireworksai.models.ts b/src/modules/llms/server/openai/models/fireworksai.models.ts index b04677e4d..5330391ed 100644 --- a/src/modules/llms/server/openai/models/fireworksai.models.ts +++ b/src/modules/llms/server/openai/models/fireworksai.models.ts @@ -74,7 +74,6 @@ export function fireworksAIModelsToModelDescriptions(wireModels: unknown): Model interfaces, // parameterSpecs: ... // maxCompletionTokens: ... - // trainingDataCutoff: ... // benchmark: ... // chatPrice, hidden: false, diff --git a/src/modules/llms/server/openai/models/localai.models.ts b/src/modules/llms/server/openai/models/localai.models.ts index e6ade0958..e4022dc41 100644 --- a/src/modules/llms/server/openai/models/localai.models.ts +++ b/src/modules/llms/server/openai/models/localai.models.ts @@ -58,7 +58,6 @@ export function localAIModelToModelDescription(modelId: string): ModelDescriptio interfaces, // parameterSpecs // maxCompletionTokens - // trainingDataCutoff // benchmark chatPrice: _knownLocalAIPrice, hidden: _hideLocalAIModels.includes(modelId), diff --git a/src/modules/llms/server/openai/models/mistral.models.ts b/src/modules/llms/server/openai/models/mistral.models.ts index 481620e3e..370e31937 100644 --- a/src/modules/llms/server/openai/models/mistral.models.ts +++ b/src/modules/llms/server/openai/models/mistral.models.ts @@ -235,7 +235,6 @@ export function mistralModels(wireModels: unknown): ModelDescriptionSchema[] { interfaces: _mistralCapabilitiesToInterfaces(capabilities, id), // parameterSpecs: ... // maxCompletionTokens: ... - // trainingDataCutoff: ... // benchmark, chatPrice, hidden: provided by extraDetails below: ...extraDetails, // Override hidden only if not explicitly set in extraDetails diff --git a/src/modules/llms/server/openai/models/openai.models.ts b/src/modules/llms/server/openai/models/openai.models.ts index 92875087c..01d04aec4 100644 --- a/src/modules/llms/server/openai/models/openai.models.ts +++ b/src/modules/llms/server/openai/models/openai.models.ts @@ -80,7 +80,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Most capable model for professional work and long-running agents. Improvements in general intelligence, long-context, agentic tool-calling, and vision.', contextWindow: 400000, maxCompletionTokens: 128000, - trainingDataCutoff: 'Aug 2025', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_Tools_WebSearch, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmVndOaiReasoningEffort52', initialValue: 'medium' /* our decision: set to medium to have thinking - clones can set to 'none' to have temperature */ }, @@ -106,7 +105,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'GPT-5.2 optimized for long-horizon, agentic coding tasks in Codex or similar environments. Supports low, medium, high, and xhigh reasoning effort settings.', contextWindow: 400000, maxCompletionTokens: 128000, - trainingDataCutoff: 'Aug 2025', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmForceNoStream' }, @@ -124,7 +122,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'GPT-5.2 model powering ChatGPT. Fast, capable for everyday work with clear improvements in info-seeking, how-tos, technical writing.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Aug 2025', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE, LLM_IF_Tools_WebSearch, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmVndOaiWebSearchContext' }, @@ -143,7 +140,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Smartest and most trustworthy option for difficult questions. Uses more compute for harder thinking on complex domains like programming.', contextWindow: 400000, maxCompletionTokens: 272000, - trainingDataCutoff: 'Aug 2025', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_MIN, LLM_IF_OAI_Reasoning, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmVndOaiReasoningEffort52Pro' }, @@ -169,7 +165,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'The best model for coding and agentic tasks with configurable reasoning effort.', contextWindow: 400000, maxCompletionTokens: 128000, - // trainingDataCutoff: 'Sep 30, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_Tools_WebSearch, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmVndOaiReasoningEffort4' }, { paramId: 'llmVndOaiWebSearchContext' }, @@ -194,7 +189,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'GPT-5.1 Instant with adaptive reasoning. More conversational with improved instruction following.', contextWindow: 128000, maxCompletionTokens: 16384, - // trainingDataCutoff: 'Oct 31, 2024', // interfaces: [LLM_IF_OAI_Responses, LLM_IF_OAI_Chat, LLM_IF_OAI_Vision, LLM_IF_OAI_PromptCaching], // no function calling or reasoning interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_Tools_WebSearch, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ @@ -214,7 +208,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Our most intelligent coding model optimized for long-horizon, agentic coding tasks.', contextWindow: 400000, maxCompletionTokens: 128000, - trainingDataCutoff: 'Sep 30, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmVndOaiReasoningEffort4' }, @@ -230,7 +223,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'A version of GPT-5.1 optimized for agentic coding tasks in Codex or similar environments.', contextWindow: 400000, maxCompletionTokens: 128000, - trainingDataCutoff: 'Sep 30, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmVndOaiReasoningEffort' }, // low, medium, high (no minimal) @@ -246,7 +238,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Smaller, faster version of GPT-5.1 Codex for efficient coding tasks.', contextWindow: 400000, maxCompletionTokens: 128000, - trainingDataCutoff: 'Sep 30, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmVndOaiReasoningEffort' }, // low, medium, high (no minimal) @@ -267,7 +258,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'The best model for coding and agentic tasks across domains.', contextWindow: 400000, maxCompletionTokens: 128000, - trainingDataCutoff: 'Sep 30, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_Tools_WebSearch, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmVndOaiReasoningEffort4' }, { paramId: 'llmVndOaiWebSearchContext' }, @@ -292,7 +282,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Version of GPT-5 that uses more compute to produce smarter and more precise responses. Designed for tough problems.', contextWindow: 400000, maxCompletionTokens: 272000, - trainingDataCutoff: 'Sep 30, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_MIN, LLM_IF_OAI_Reasoning, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [{ paramId: 'llmForceNoStream' }], // reasoning effort is fixed at 'high' chatPrice: { input: 15, output: 120 }, @@ -311,7 +300,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'GPT-5 model used in ChatGPT. Points to the GPT-5 snapshot currently used in ChatGPT.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Sep 29, 2024', interfaces: [LLM_IF_OAI_Responses, LLM_IF_OAI_Chat, LLM_IF_OAI_Vision, LLM_IF_OAI_PromptCaching], // no function calling or reasoning chatPrice: { input: 1.25, cache: { cType: 'oai-ac', read: 0.125 }, output: 10 }, benchmark: { cbaElo: 1430 }, // gpt-5-chat @@ -324,7 +312,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'A version of GPT-5 optimized for agentic coding in Codex.', contextWindow: 400000, maxCompletionTokens: 128000, - trainingDataCutoff: 'Sep 30, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_Tools_WebSearch, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [ { paramId: 'llmVndOaiReasoningEffort' }, // works @@ -343,7 +330,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Updated web search model in Chat Completions API. 60% cheaper with domain filtering support.', contextWindow: 400000, maxCompletionTokens: 100000, - trainingDataCutoff: 'Sep 30, 2024', interfaces: [...IFS_CHAT_MIN, LLM_IF_Tools_WebSearch], parameterSpecs: [{ paramId: 'llmVndOaiWebSearchContext', initialValue: 'medium' }], // Search enabled by default chatPrice: { input: 1.25, cache: { cType: 'oai-ac', read: 0.125 }, output: 10 }, @@ -362,7 +348,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'A faster, more cost-efficient version of GPT-5 for well-defined tasks.', contextWindow: 400000, maxCompletionTokens: 128000, - trainingDataCutoff: 'May 30, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_Tools_WebSearch, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [{ paramId: 'llmVndOaiReasoningEffort4' }, { paramId: 'llmVndOaiWebSearchContext' }, { paramId: 'llmVndOaiVerbosity' }, { paramId: 'llmVndOaiImageGeneration' }, { paramId: 'llmForceNoStream' }], chatPrice: { input: 0.25, cache: { cType: 'oai-ac', read: 0.025 }, output: 2 }, @@ -381,7 +366,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Fastest, most cost-efficient version of GPT-5 for summarization and classification tasks.', contextWindow: 400000, maxCompletionTokens: 128000, - trainingDataCutoff: 'May 30, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [{ paramId: 'llmVndOaiReasoningEffort4' }, { paramId: 'llmVndOaiVerbosity' }, { paramId: 'llmVndOaiImageGeneration' }], chatPrice: { input: 0.05, cache: { cType: 'oai-ac', read: 0.005 }, output: 0.4 }, @@ -419,7 +403,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Specialized model for computer use tool. Optimized for computer interaction capabilities.', contextWindow: 8192, maxCompletionTokens: 1024, - trainingDataCutoff: 'Sep 30, 2023', interfaces: [LLM_IF_OAI_Responses, LLM_IF_OAI_Chat, LLM_IF_OAI_Fn, LLM_IF_HOTFIX_NoTemperature], chatPrice: { input: 3, output: 12 }, isPreview: true, @@ -436,7 +419,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Fast reasoning model optimized for the Codex CLI. A fine-tuned version of o4-mini for low-latency code Q&A and editing.', contextWindow: 200000, maxCompletionTokens: 100000, - trainingDataCutoff: 'May 31, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [{ paramId: 'llmVndOaiReasoningEffort' }], chatPrice: { input: 1.5, cache: { cType: 'oai-ac', read: 0.375 }, output: 6 }, @@ -453,7 +435,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Faster, more affordable deep research model for complex, multi-step research tasks.', contextWindow: 200000, maxCompletionTokens: 100000, - trainingDataCutoff: 'May 31, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON], parameterSpecs: PS_DEEP_RESEARCH, chatPrice: { input: 2, cache: { cType: 'oai-ac', read: 0.5 }, output: 8 }, @@ -471,7 +452,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Latest o4-mini model. Optimized for fast, effective reasoning with exceptionally efficient performance in coding and visual tasks.', contextWindow: 200000, maxCompletionTokens: 100000, - trainingDataCutoff: 'May 31, 2024', interfaces: IFS_CHAT_CACHE_REASON, parameterSpecs: [{ paramId: 'llmVndOaiReasoningEffort' }], chatPrice: { input: 1.1, cache: { cType: 'oai-ac', read: 0.275 }, output: 4.4 }, @@ -490,7 +470,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Our most powerful deep research model for complex, multi-step research tasks.', contextWindow: 200000, maxCompletionTokens: 100000, - trainingDataCutoff: 'May 31, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_CACHE_REASON], parameterSpecs: PS_DEEP_RESEARCH, chatPrice: { input: 10, cache: { cType: 'oai-ac', read: 2.5 }, output: 40 }, @@ -508,7 +487,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Version of o3 with more compute for better responses. Provides consistently better answers for complex tasks.', contextWindow: 200000, maxCompletionTokens: 100000, - trainingDataCutoff: 'May 31, 2024', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_MIN, LLM_IF_OAI_Reasoning, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [{ paramId: 'llmVndOaiReasoningEffort' }, { paramId: 'llmForceNoStream' }], chatPrice: { input: 20, output: 80 }, @@ -527,7 +505,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'A well-rounded and powerful model across domains. Sets a new standard for math, science, coding, and visual reasoning tasks.', contextWindow: 200000, maxCompletionTokens: 100000, - trainingDataCutoff: 'May 31, 2024', interfaces: IFS_CHAT_CACHE_REASON, parameterSpecs: [{ paramId: 'llmVndOaiReasoningEffort' }, { paramId: 'llmForceNoStream' }], chatPrice: { input: 2, cache: { cType: 'oai-ac', read: 0.5 }, output: 8 }, @@ -546,7 +523,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Latest o3-mini model snapshot. High intelligence at the same cost and latency targets of o1-mini. Excels at science, math, and coding tasks.', contextWindow: 200000, maxCompletionTokens: 100000, - trainingDataCutoff: 'Oct 2023', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn, LLM_IF_OAI_Json, LLM_IF_OAI_PromptCaching, LLM_IF_OAI_Reasoning, LLM_IF_HOTFIX_StripImages], parameterSpecs: [{ paramId: 'llmVndOaiReasoningEffort' }], chatPrice: { input: 1.1, cache: { cType: 'oai-ac', read: 0.55 }, output: 4.4 }, @@ -566,7 +542,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'A version of o1 with more compute for better responses. Provides consistently better answers for complex tasks.', contextWindow: 200000, maxCompletionTokens: 100000, - trainingDataCutoff: 'Sep 30, 2023', interfaces: [LLM_IF_OAI_Responses, ...IFS_CHAT_MIN, LLM_IF_OAI_Reasoning, LLM_IF_HOTFIX_NoTemperature], parameterSpecs: [{ paramId: 'llmVndOaiReasoningEffort' }, { paramId: 'llmForceNoStream' }], chatPrice: { input: 150, output: 600 }, @@ -585,7 +560,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Previous full o-series reasoning model.', contextWindow: 200000, maxCompletionTokens: 100000, - trainingDataCutoff: 'Oct 2023', interfaces: IFS_CHAT_CACHE_REASON, parameterSpecs: [{ paramId: 'llmVndOaiReasoningEffort' }, { paramId: 'llmVndOaiRestoreMarkdown' }], chatPrice: { input: 15, cache: { cType: 'oai-ac', read: 7.5 }, output: 60 }, @@ -607,7 +581,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Flagship GPT model for complex tasks. Major improvements on coding, instruction following, and long context with 1M token context window.', contextWindow: 1047576, maxCompletionTokens: 32768, - trainingDataCutoff: 'May 31, 2024', interfaces: IFS_CHAT_CACHE, chatPrice: { input: 2, cache: { cType: 'oai-ac', read: 0.5 }, output: 8 }, benchmark: { cbaElo: 1409 }, // gpt-4.1-2025-04-14 @@ -625,7 +598,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Balanced for intelligence, speed, and cost. Matches or exceeds GPT-4o in intelligence while reducing latency by nearly half and cost by 83%.', contextWindow: 1047576, maxCompletionTokens: 32768, - trainingDataCutoff: 'May 31, 2024', interfaces: IFS_CHAT_CACHE, chatPrice: { input: 0.4, cache: { cType: 'oai-ac', read: 0.1 }, output: 1.6 }, benchmark: { cbaElo: 1377 }, // gpt-4.1-mini-2025-04-14 @@ -643,7 +615,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Fastest, most cost-effective GPT 4.1 model. Delivers exceptional performance with low latency, ideal for tasks like classification or autocompletion.', contextWindow: 1047576, maxCompletionTokens: 32768, - trainingDataCutoff: 'May 31, 2024', interfaces: IFS_CHAT_CACHE, chatPrice: { input: 0.1, cache: { cType: 'oai-ac', read: 0.025 }, output: 0.4 }, benchmark: { cbaElo: 1320 }, // gpt-4.1-nano-2025-04-14 @@ -710,7 +681,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Snapshot of gpt-4o from November 20th, 2024.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Oct 2023', interfaces: IFS_CHAT_CACHE, chatPrice: { input: 2.5, cache: { cType: 'oai-ac', read: 1.25 }, output: 10 }, benchmark: { cbaElo: 1265 + 1 }, // not reported; using gpt-4o-2024-08-06 + 1 @@ -722,7 +692,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Snapshot that supports Structured Outputs. gpt-4o currently points to this version.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Oct 2023', interfaces: IFS_CHAT_CACHE, chatPrice: { input: 2.5, cache: { cType: 'oai-ac', read: 1.25 }, output: 10 }, benchmark: { cbaElo: 1333 }, // GPT-4o (08/06) @@ -734,7 +703,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Original gpt-4o snapshot from May 13, 2024.', contextWindow: 128000, maxCompletionTokens: 4096, - trainingDataCutoff: 'Oct 2023', interfaces: IFS_CHAT_MIN, chatPrice: { input: 5, output: 15 }, benchmark: { cbaElo: 1344 }, // gpt-4o-2024-05-13 @@ -750,7 +718,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'The chatgpt-4o-latest model version continuously points to the version of GPT-4o used in ChatGPT, and is updated frequently.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Oct 2023', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Vision, LLM_IF_OAI_Json], // does not support Tools chatPrice: { input: 5, output: 15 }, benchmark: { cbaElo: 1441 }, // chatgpt-4o-latest-20250326 @@ -765,7 +732,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Latest snapshot of the GPT-4o model optimized for web search capabilities.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Sep 30, 2023', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Json, LLM_IF_Tools_WebSearch, LLM_IF_HOTFIX_NoTemperature], // NOTE: 2025-03-15: confirmed on 'playground' that this model does not support images parameterSpecs: [{ paramId: 'llmVndOaiWebSearchContext' }, { paramId: 'llmVndOaiWebSearchGeolocation' }], chatPrice: { input: 2.5, output: 10 }, @@ -786,7 +752,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Latest snapshot for the Audio API model.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Oct 2023', interfaces: IFS_GPT_AUDIO, chatPrice: { input: 2.5, output: 10 /* AUDIO PRICING UNSUPPORTED 40/80 */ }, // benchmarks don't apply to audio models @@ -799,7 +764,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Snapshot for the Audio API model.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Oct 2023', interfaces: IFS_GPT_AUDIO, chatPrice: { input: 2.5, output: 10 /* AUDIO PRICING UNSUPPORTED 40/80 */ }, // benchmarks don't apply to audio models @@ -818,7 +782,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Affordable model for fast, lightweight tasks. GPT-4o Mini is cheaper and more capable than GPT-3.5 Turbo.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Oct 2023', interfaces: IFS_CHAT_CACHE, chatPrice: { input: 0.15, cache: { cType: 'oai-ac', read: 0.075 }, output: 0.6 }, benchmark: { cbaElo: 1316 }, // GPT-4o-mini (07/18) @@ -835,7 +798,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Snapshot for the Audio API model.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Oct 2023', interfaces: IFS_GPT_AUDIO, chatPrice: { input: 0.15, output: 0.6 /* AUDIO PRICING UNSUPPORTED 10/20 */ }, // benchmarks don't apply to audio models @@ -854,7 +816,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'Latest snapshot of the GPT-4o Mini model optimized for web search capabilities.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Sep 30, 2023', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Json, LLM_IF_Tools_WebSearch, LLM_IF_HOTFIX_NoTemperature], // NOTE: this support function calling, but only its own, not a Custom Function parameterSpecs: [{ paramId: 'llmVndOaiWebSearchContext' }, { paramId: 'llmVndOaiWebSearchGeolocation' }], chatPrice: { input: 0.15, output: 0.6 }, @@ -875,7 +836,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'GPT-4 Turbo with Vision model. Vision requests can now use JSON mode and function calling. gpt-4-turbo currently points to this version.', contextWindow: 128000, maxCompletionTokens: 4096, - trainingDataCutoff: 'Dec 2023', interfaces: IFS_CHAT_MIN, chatPrice: { input: 10, output: 30 }, benchmark: { cbaElo: 1324 }, // gpt-4-turbo-2024-04-09 @@ -892,7 +852,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'GPT-4 Turbo preview model intended to reduce cases of "laziness" where the model doesn\'t complete a task.', contextWindow: 128000, maxCompletionTokens: 4096, - trainingDataCutoff: 'Dec 2023', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn, LLM_IF_OAI_Json], chatPrice: { input: 10, output: 30 }, benchmark: { cbaElo: 1315 }, // gpt-4-0125-preview @@ -904,7 +863,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'GPT-4 Turbo preview model featuring improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more.', contextWindow: 128000, maxCompletionTokens: 4096, - trainingDataCutoff: 'Apr 2023', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn, LLM_IF_OAI_Json], chatPrice: { input: 10, output: 30 }, benchmark: { cbaElo: 1315 }, // gpt-4-1106-preview @@ -923,7 +881,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ hidden: true, // OLD description: 'Snapshot of gpt-4 from June 13th 2023 with improved function calling support. Data up to Sep 2021.', contextWindow: 8192, - trainingDataCutoff: 'Sep 2021', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn], chatPrice: { input: 30, output: 60 }, benchmark: { cbaElo: 1163 }, @@ -935,7 +892,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ hidden: true, // OLD description: 'Snapshot of gpt-4 from March 14th 2023 with function calling data. Data up to Sep 2021.', contextWindow: 8192, - trainingDataCutoff: 'Sep 2021', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn], chatPrice: { input: 30, output: 60 }, benchmark: { cbaElo: 1186 }, @@ -959,7 +915,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'The latest GPT-3.5 Turbo model with higher accuracy at responding in requested formats and a fix for a bug which caused a text encoding issue for non-English language function calls.', contextWindow: 16385, maxCompletionTokens: 4096, - trainingDataCutoff: 'Sep 2021', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn], chatPrice: { input: 0.5, output: 1.5 }, benchmark: { cbaElo: 1106 }, @@ -971,7 +926,6 @@ export const _knownOpenAIChatModels: ManualMappings = [ description: 'GPT-3.5 Turbo model with improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more.', contextWindow: 16385, maxCompletionTokens: 4096, - trainingDataCutoff: 'Sep 2021', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn], chatPrice: { input: 1, output: 2 }, benchmark: { cbaElo: 1068 }, diff --git a/src/modules/llms/server/openai/models/openpipe.models.ts b/src/modules/llms/server/openai/models/openpipe.models.ts index 5cac4c340..caf7b19ad 100644 --- a/src/modules/llms/server/openai/models/openpipe.models.ts +++ b/src/modules/llms/server/openai/models/openpipe.models.ts @@ -15,7 +15,6 @@ const _knownOpenPipeChatModels: ModelDescriptionSchema[] = [ description: 'Affordable model for fast, lightweight tasks. GPT-4o mini is cheaper and more capable than GPT-3.5 Turbo.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Oct 2023', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Vision, LLM_IF_OAI_Fn, LLM_IF_OAI_Json], chatPrice: _knownOpenAIChatModels.find(m => m.idPrefix === 'gpt-4o-mini-2024-07-18')?.chatPrice, benchmark: { cbaMmlu: 82.0 }, @@ -26,7 +25,6 @@ const _knownOpenPipeChatModels: ModelDescriptionSchema[] = [ description: 'Advanced, multimodal flagship model that\'s cheaper and faster than GPT-4 Turbo.', contextWindow: 128000, maxCompletionTokens: 16384, - trainingDataCutoff: 'Oct 2023', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Vision, LLM_IF_OAI_Fn, LLM_IF_OAI_Json], chatPrice: _knownOpenAIChatModels.find(m => m.idPrefix === 'gpt-4o-2024-08-06')?.chatPrice, benchmark: { cbaElo: 1287 }, @@ -37,7 +35,6 @@ const _knownOpenPipeChatModels: ModelDescriptionSchema[] = [ description: 'The latest GPT-3.5 Turbo model with higher accuracy at responding in requested formats', contextWindow: 16385, maxCompletionTokens: 4096, - trainingDataCutoff: 'Sep 2021', interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn], chatPrice: _knownOpenAIChatModels.find(m => m.idPrefix === 'gpt-3.5-turbo-0125')?.chatPrice, benchmark: { cbaElo: 1105 }, diff --git a/src/modules/llms/server/openai/models/openrouter.models.ts b/src/modules/llms/server/openai/models/openrouter.models.ts index 0897af537..719f364e1 100644 --- a/src/modules/llms/server/openai/models/openrouter.models.ts +++ b/src/modules/llms/server/openai/models/openrouter.models.ts @@ -196,7 +196,6 @@ export function openRouterModelToModelDescription(wireModel: object): ModelDescr description: model.description?.length > 280 ? model.description.slice(0, 277) + '...' : model.description, contextWindow, maxCompletionTokens, - // trainingDataCutoff: ... interfaces, // benchmark: ... chatPrice, diff --git a/src/modules/llms/server/openai/models/together.models.ts b/src/modules/llms/server/openai/models/together.models.ts index 94a1f7036..ac86db41c 100644 --- a/src/modules/llms/server/openai/models/together.models.ts +++ b/src/modules/llms/server/openai/models/together.models.ts @@ -82,7 +82,6 @@ export function togetherAIModelsToModelDescriptions(wireModels: unknown): ModelD interfaces, // parameterSpecs: ... // maxCompletionTokens: ... - // trainingDataCutoff: ... // benchmark: ... chatPrice, hidden: false, diff --git a/src/modules/persona/pmix/pmix.parameters.ts b/src/modules/persona/pmix/pmix.parameters.ts index 87515d80a..3351ddafb 100644 --- a/src/modules/persona/pmix/pmix.parameters.ts +++ b/src/modules/persona/pmix/pmix.parameters.ts @@ -12,7 +12,6 @@ */ import type { DLLMId } from '~/common/stores/llms/llms.types'; -import { findLLMOrThrow } from '~/common/stores/llms/store-llms'; /// Types @@ -132,22 +131,6 @@ Choose the most suitable chart type based on the data and context. Include only // Model Capabilities - '{{LLM.Cutoff}}': { - scope: 'model', - description: 'Model knowledge cutoff date', - dependencies: { assistantLlmId: true }, - wholeLine: true, - replace: ({ assistantLlmId }) => { - try { - if (assistantLlmId) - return findLLMOrThrow(assistantLlmId).trainingDataCutoff || null; - } catch (e) { - // ignore... - } - return null; - }, - }, - '{{LLM.LowRL:...}}': { scope: 'model', description: 'Conditional content for non-reasoning models', diff --git a/src/modules/persona/pmix/pmix.ts b/src/modules/persona/pmix/pmix.ts index 46cc1610b..8c711cdcf 100644 --- a/src/modules/persona/pmix/pmix.ts +++ b/src/modules/persona/pmix/pmix.ts @@ -67,6 +67,10 @@ export function replacePromptVariables(template: string, context: PPromptMixerCo for (const [placeholder, replacement] of Object.entries(context.customFields)) mixed = mixed.replaceAll(placeholder, replacement); + // Remove deprecated variables (whole line removal) + // - {{LLM.Cutoff}} - deprecated on 2026-01-29; not used, maybe back to gpt-4, but then we could just update those prompts + mixed = mixed.replaceAll(/.*\{\{LLM\.Cutoff}}.*\n/g, ''); + // At most leave 2 newlines in a row mixed = mixed.replace(/\n{3,}/g, '\n\n');